ANIMACY EFFECTS IN INTRANSITIVE VERBS Running Head: Animacy effects in intransitive verbs Animacy effects on the processing of intransitive verbs: an eye-tracking study Mirta Vernice1 Antonella Sorace2 1Department of Psychology, University of Milano-Bicocca, Italy 2 School of Philosophy, Psychology and Language Sciences, University of Edinburgh, Edinburgh, UK Contact: Mirta Vernice Department of Psychology, U6 Building, Piazza dell'Ateneo Nuovo 1, 20126, Milano, Italy E-mail: [email protected] 1 ANIMACY EFFECTS IN INTRANSITIVE VERBS Abstract This paper tested an assumption of the gradient model of split intransitivity put forward by Sorace (‘Split Intransitivity Hierarchy’ (SIH), 2000, 2004), namely that agentivity is a fundamental feature for unergatives but not for unaccusatives. According to this hypothesis, the animacy of the verb’s argument should affect the processing of unergative verbs to a greater extent than unaccusative verbs. By using eye-tracking methodology we monitored the on-line processing and integration costs of the animacy of the verb’s argument in intransitive verbs. We observed that inanimate subjects caused longer reading times only for unergative verbs, whereas the animacy of the verb’s argument did not influence the pattern of results for unaccusatives. In addition, the unergative verb data directly support the existence of gradient effects on the processing of the subject argument. 2 ANIMACY EFFECTS IN INTRANSITIVE VERBS Animacy effects on the processing of intransitive verbs: an eye-tracking study The Unaccusative Hypothesis (Burzio, 1986) holds that, across languages, intransitive verbs can be divided into two classes, i.e., unaccusatives (e.g., to arrive) and unergatives (e.g., to work). Much linguistic research has differentiated these two verb classes with respect to their inherent syntactic properties (e.g., Burzio, 1986; Ackema and Sorace, 2017); however, alongside the syntactic distinction, the unaccusative/unergative split is sensitive to the semantic characteristics of the verb and of its argument (cf., Sorace, 2000; 2004). In the current study we tested whether a manipulation of a semantic feature of the verb’s argument, namely animacy, affects the processing of Italian intransitive verbs. A relevant difference between unaccusatives and unergatives relies in their semantic underpinnings. Semantically, unaccusative verbs involve actions that typically involve a change of state (e.g., die, disappear), whereas unergatives typically denote controlled motional processes (i.e., work) (Sorace, 2000). There are crucial semantic implications in terms of the underlying thematic properties of the argument of the verb: unaccusatives require a non-agentive argument, whereas unergatives involve actions that are performed by a proto-typically agentive argument (cf. Dowty, 1991; Van Valin, 1990). Experimental evidence based on transitive (agentive) events suggests that animate agents are much preferred over inanimate ones because they are more prototypical and thus more accessible (Keenan & Comrie, 1977; see also Bock & Warren, 1985; Bock, 1986; Bock & Loebell, 1990). Additionally, there is a bulk of experimental evidence indicating that under some conditions animate agents are more likely to be assigned to higher grammatical functions (Comrie, 1989; Dowty, 1991; Hopper and Thompson, 1980; Langacker 1991) with this preference being overcome only when agents are inanimate and patients animate (Bock, 1986; Bock & Loebell, 1990). The 3 ANIMACY EFFECTS IN INTRANSITIVE VERBS tendency to prefer animate agentive entities (in transitive events) in early sentence positions or in higher grammatical roles has been reported in production (e.g., Prat-Sala & Branigan, 2000), and in comprehension too (Kuperberg et al. 2007; Townsend and Bever, 2001). For instance, Weckerly & Kutas (1999) found that sentence-initial inanimate entities occurring in the grammatical role of subject caused significant processing costs and a specific N400-P600 ERP pattern in relative clauses comprehension. In an EEG study, Malaia et al. (2015) tested the processing of the subject noun animacy in the comprehension of relative clauses. Again, inanimate subjects elicited an anterior negative shift and caused a slowdown in sentence comprehension. In the above mentioned cases, animate agents are preferred when they occur in subject/first position. However, note that this preference is far from being deterministic: it is important to consider a number of additional constraints such as, for instance, the pragmatic function the argument plays in the discourse (whether it is the most emphasised element in a sentence; cf. Gordon, Grosz, & Gilliom, 1993), or the thematic fit of the argument with respect to the verb (McRae, Ferretti & Amyote, 1997; McRae, Spivey-Knowlton and Tanenhaus, 1998; Ferretti et al., 2001). Therefore one might conclude that, under some constraints, an inanimate agent might well be placed as the subject of an active sentence. To our knowledge, only few studies have focused on the processing of the subject of intransitive verbs and its effects on split intransitivity. By means of a cross-modal lexical priming technique, Friedmann et al. (2008) tested whether the subject was reactivated more often during the online processing of a sentence involving an unergative or an unaccusative verb. The study revealed that only subjects of unaccusatives were reactivated after the verb, whereas subjects of unergatives were not. One possible explanation for such finding is that, as the argument of unergatives is syntactically a subject, participants did not need to reactivate its trace; in contrast, in unaccusatives, the subject’s trace was less salient, and therefore had to be reactivated. All in all, this finding offers support to the view that the single argument of unaccusative and unergative verbs might involve differential processing strategies for the speaker. 4 ANIMACY EFFECTS IN INTRANSITIVE VERBS According to the literature, however, semantic differences are strictly intertwined with the syntactic distinctions between these verbs classes, suggesting that a more exhaustive explanation of the phenomena at stake here has to be found at the syntax-semantics interface (Levin and Rappaport Hovav, 1995). For this reason, let us briefly summarize the classic syntactic explanation of split intransitivity. According to the Unaccusative Hypothesis (e.g., Burzio, 1986), single arguments of unaccusative verbs are derived by an operation that eliminates the subject of a transitive verb (Reinhart, 2002), leaving the direct object as the only argument of the verb, in the canonical object (post-verbal) position; in contrast, the subject of unergative verbs is equivalent to (and behaves like) the subject of a transitive verb. Thus, the single argument of an unaccusative verb is syntactically comparable to the object of a transitive verb, while the single argument of an unergative verb is comparable to a subject. Further evidence for the object status of the unaccusative argument is provided by the NE- cliticization test in Italian (Sorace, 1995), which works for unaccusatives (1) as it does for the object of transitives (2), but not for unergatives (3). (1) Ne arriva uno. Of them arrives one. ‘One of them is arriving.’ (2) Ne ho preso uno. Of them I took one. ‘I took one of them.’ (3) *Ne chiacchiera uno. Of them chats one. ‘One of them is chatting.’ 5 ANIMACY EFFECTS IN INTRANSITIVE VERBS The choice of perfective auxiliary is also regarded as a syntactic diagnostics of the unaccusativity/unergativity split. For example, in several European languages, among which Italian, unaccusative verbs generally select ‘be’ as a perfective auxiliary while unergative verbs select ‘have’. In our examples below, (4) involves an unaccusative verb, i.e., arrivare (‘to come’), and thus selects essere (‘be’) as auxiliary, whereas the auxiliary avere (‘have’) is ungrammatical. On the other hand, the unergative verb chiacchierare (‘to chat’) can take only avere. (4) Il ragazzo è / *ha arrivato in ritardo. ‘The boy is / *has arrived late.’ (5) Il ragazzo *è / ha chiacchierato con gli amici. ‘The boy *is / has chatted with friends.’ However, as noted by Sorace (2000), the choice of the auxiliary could not be reduced to a categorical choice (cf. Burzio, 1986). In this regard, she observed that auxiliary selection is crucially determined by the lexical-semantic properties of the verb. That is, the choice of the auxiliary appears to be modulated in a gradient fashion by aspectual features (telicity/atelicity) of the verb and of the predicate in which the verb appears, as well as by the degree of agentivity of the argument of the verb. She thus proposed a hierarchy of semantic verb classes, termed ‘Auxiliary Selection Hierarchy’ (ASH) (reported below in 6), that captures the gradient likelihood of an intransitive verb to select the auxiliary ‘to have’ or ‘to be’. (6) Auxiliary Selection Hierarchy (ASH) change of location verbs select BE change of state continuation of state existence of state uncontrolled process controlled process (motional) controlled process (non-motional) verbs select HAVE 6 ANIMACY EFFECTS IN INTRANSITIVE VERBS According to the ASH, verbs involving change of location or change of state are ‘core unaccusatives’, lexically denote telic change and categorically select ‘be’ as auxiliary; verbs involving non-motional controlled processes are ‘core unergatives’, denote an agentive activity and categorically take ‘have’. Thus, at the ‘be end’ of the ASH